in the Pacific, S^c. 



143 



H. the low, sandy island between the entrances, which are marked by the 

 dotted line. 



The four trees represent the situation of as many groves of cocoa-nut. The 

 dark dotted patches in the lagoon, which should be much more numerous, are 

 sand banks and coral shallows; and the black spots on the eastern shore, denote 

 large fragments of reef-rock thrown up by the breakers. 



At Carlshoff Island, in about 15° 30' S. lat. and 145° 30' W. 

 long., near the north-west side of the lagoon, and a short 

 quarter of a mile from the sea beach, is a pool of tolerably 

 fresh water, between fifty and sixty yards round, and five 

 feet deep, which appears to have been formed like the dry 

 hollows at Christmas Island. 



But in nothing perhaps throughout the coral seas, are the 

 proofs of re-elevation more conspicuous than in the large 

 tabular masses of reef-rock which have been spoken of as 

 lining the weather shores of many Paumotus. At Serle, Vin- 

 cennes, (a few miles West of Raraka,) King's, Carlshofi", 

 Honden, Tooa, Aitoho, and Dean's or Prince of Wales' islands, 

 these constitute one of the most prominent features. At the 

 last named, they are strewn along the coast for upwards of 

 thirty miles, and some of the masses, as nearly as I could 

 estimate, were a hundred feet long. It is not improbable that 

 the whole coast here may be a reef raised three or four feet. 

 The enormous size of the rocks in some instances, renders it 

 almost incredible that any surf could have been sufficiently 

 powerful, to tear from the reef and remove them to their pres- 

 ent situation. 



Not having landed here, but only seen these ledges through a 

 glass while coasting the island at a mile's distance, I cannot 

 of course speak positively on this question, or as to their exact 

 size ; but as regards the latter, any person who has made the 

 experiment, will admit that in viewing objects on a beach in 

 this manner, their magnitude is more apt to be under than 

 over-estimated. In most cases, however, there is no doubt 

 but that these blocks are erratic, and originally constituted 



